ANIMAL BEHAVIOUR, 2003, 66, 169–174doi:10.1006/anbe.2003.2191

Environmental and genetic background of tiptoe-initiatingbehaviour in the dwarfspider Erigone atra

DRIES BONTE*, ISRA DEBLAUWE* & JEAN-PIERRE MAELFAIT*†

*Department of Biology, University of Ghent†Institute of Nature Conservation, Brussels

(Received 13 February 2002; initial acceptance 1 May 2002;final acceptance 24 November 2002; MS. number: 7240R)

Ballooning is a widespread behavioural trait in invertebrates, which enhances passive aerial dispersal. Weinvestigated the influence of common lineage (family effect) and postnatal environmental conditions onlatency to initiate preballooning tiptoe behaviour (ballooning latency) in the dwarfspider Erigone atra(Blackwall, 1841). The ballooning latency of individual spiders was tested in a specially designed testchamber. In a 3-week experiment, in which the spiders were fed only during the first day of each week,ballooning latency had low repeatability at weekly intervals. Ballooning latency declined with increasingfood deprivation during the first week but not during the second and the third weeks. At intervals of lessthan 1 h, however, ballooning latency showed high repeatability. We also investigated whether maternaland postnatal environmental conditions (i.e. during juvenile development) influence phenotypicvariation in ballooning latency. We reared offspring of several families under two feeding and twotemperature conditions. Environmental conditions explained more variation in ballooning latency thanfamily. Ballooning latency was lower in spiders reared at 20�C than in those reared at 15�C. In addition,spiderlings fed four prey per 3 days were faster ballooners than those fed only four prey per week. Aninteraction between factors was present, indicating the existence of different reaction norms between thetwo environmental conditions. The expression of ballooning latency behaviour thus strongly dependson current nutrition, feeding history and the feeding and temperature conditions during juveniledevelopment.

2003 Published by Elsevier Science Ltd on behalf of The Association for the Study of Animal Behaviour.

Correspondence: D. Bonte and I. Deblauwe, Department ofBiology, Research Unit of Terrestrial Ecology, Ghent University,K. L. Ledeganckstraat 35, B-9000 Ghent, Belgium (email: dries.bonte@agent.be). J.-P. Maelfait is at the Institute of NatureConservation, Kliniekstraat 25, B-1070 Brussels, Belgium.

Aerial dispersal is common among terrestrial insects andspiders. In the latter, this mode of dispersal is known asballooning. The spider climbs to an elevated startingpoint, stretches its legs and raises its abdomen (tiptoebehaviour); it then produces long silk threads and isfinally carried up by the air current (Jones 1996a, b).Aerial dispersal is effective for both distant migration(possibly affecting gene flow) and local dispersal (Thomas1996), especially in ephemeral but highly productivehabitats where rapid movement under unfavourable con-ditions is important for survival (Legel & Van Wingerden1980; Van Wingerden 1980; Morse 1993; Duffey 1998).Tiptoe behaviour is typical of aerial dispersal in spiders ofthe Linyphiidae and juvenile Lycosidae and can be easilyobserved under laboratory conditions (Weyman 1993).Erigone atra, one of the commonest spiders of openhabitats in the temperate regions of the northern hemi-

0003–3472/03/$30.00/0 2003 Published by Elsevie169

sphere, has a strongly developed ballooning behaviour.This species is often one of the first pioneers onregularly disturbed, especially agricultural, land (De Keer& Maelfait 1988a, b).

Ballooning in spiders has mainly been studied in fieldconditions, especially within the framework of timing ofdispersal (Dean & Sterling 1985; Greenstone et al. 1987;Blandenier & Furst 1998; Bonte et al. 1998) and inrelation to spider densities and activities in the field(Greenstone et al. 1985; Weyman et al. 1995). Laboratorystudies have largely focused on the relation betweenaeronautic behaviour and meteorological conditions suchas wind velocity and temperature (Van Wingerden &Vugts 1974; Vugts & Van Wingerden 1976; Greenstone1990), and on the importance of crowding and foodshortage as a trigger for this behaviour (Legel & VanWingerden 1980). Ballooning is triggered by acute foodshortage (Weyman et al. 1994), low wind velocities(<3 m/s) and a high temperature inversion during themorning (Richter 1970; Van Wingerden & Vugts 1974;Vugts & Van Wingerden 1976; Greenstone 1990; Duffey1998), and its frequency differs between species (Richter

r Ltd on behalf of The Association for the Study of Animal Behaviour.

170 ANIMAL BEHAVIOUR, 66, 1

1970; Van Wingerden 1980; Greenstone 1982; Miller1984). Differences between populations, however, havenot been unambiguously demonstrated (Richter 1971).

Little experimental work has been carried out on factorsunderlying the initiation of ballooning behaviour, oron genetic variance underlying phenotypic variation(Weyman 1993). Several studies have shown that wingdimorphism in insects is under polygenic control(Jarvinen & Vepsalainen 1976; Roff 1986; Kaitala 1988;Desender 1989) and Li & Margolies (1993) found additivegenetic variation for dispersal behaviour in the two-spotted spider mite, Tetranychus urticae. In the latterstudy, however, nonadditive genetic variation andmaternal effects or common environmental variationexplained the largest part of the phenotypic variation,and no covariation with fecundity and sex ratio wasfound. A major problem exists in estimating heritabilityof behavioural traits because of their low level of repeat-ability, caused by difficulties in keeping experimentalconditions or the individual’s internal state constant(Kecic & Marinkov 1974).

We investigated (1) the repeatability of preballooningbehaviour, (2) the extent to which variation is deter-mined by environmental circumstances during juveniledevelopment, and (3) the extent to which a family effectis present. Such factors may contribute to similarity inoffspring dispersal behaviour, as demonstrated by Massot& Clobert (2000). Family effects include maternal effectsand possible additive heritabilities. Postnatal effectsinclude common environmental conditions and (social)interactions between siblings. Relations between off-spring dispersal and maternal and offspring characteris-tics can result from both genetic and environmentalfactors (Kirkpatrick & Lande 1989) and are very informa-tive about the nature of dispersal. Our aim was thereforeto investigate whether family effects or postnatalenvironmental conditions influence preballooningbehaviour (i.e. the latency to show tiptoe behaviour).

Although behaviours such as climbing vegetation andtaking off are also important factors in aerial dispersal(Weyman 1993), studies on the latency to initiate tiptoebehaviour are needed to understand variation in thespider’s motivation to disperse by ballooning undersuitable meteorological conditions. Fast-reacting individ-uals under laboratory conditions are more likely to bal-loon under variable weather (wind velocity) conditionsand under sudden unsuitable environmental conditionsin which spiders have to escape from their originalhabitat.

METHODS

General Methods

We tested the propensity to express tiptoe behaviourunder constant climatological conditions, using a testchamber based on the recommendations of Legel & VanWingerden (1980). Spiders were placed on a plastic framein an air tunnel with upward currents at a temperature of30�2�C, aerial humidity of 40–50% and a light breeze,velocity 1–1.2 m/s. To prevent the spiders from escaping,

we placed the frame in a bath of water. Becauseall individuals expressed the tiptoe behaviour within10 min, we measured the latency (to the nearest second)of the spider to initiate this behaviour (ballooninglatency) from the onset of wind turbulence. To be valid, aminimum of 3 s of constant tiptoe behaviour needed tobe observed. With the exception of the 3-week repeat-ability experiment (see below), all individuals were tested1 day after moulting to the adult stage, to minimizevariation in internal state. Spiders were not acclimatizedto the climate conditions in the test chamber.

We collected gravid female spiders randomly at the endof September 1999 in grass tussocks in a pasture of theWesthoek dune reserve in De Panne, Belgium. Becausethe species is abundant throughout the year, we did notselect individuals that were dependent on ballooning tocolonize sites. We used the adult instars of the females’offspring for laboratory experiments. Spiderlings werereared in petri dishes (diameter 5 cm) on plaster of Parisunder a relative humidity close to saturation and a light:dark cycle of 16:8 h. The collembolan Isotoma viridis wasused as prey. Because we wanted to investigate howdifferent environmental conditions influence ballooningbehaviour, we imposed four different rearing treatmentsby combining two temperature regimes (low: 15�C; high:20�C) and two diet regimes (well fed: four prey/3 days;poorly fed: four prey/week). All prey offered wereconsumed, and mortality was low (5.2%).

Latency data were ln transformed and did not differfrom a normal distribution (Kolmogorov–Smirnov test:NS).

Repeatability and Acute StressTo measure repeatability over short intervals, we

tested 38 randomly selected individuals twice within 1 h.Individual ballooning latencies were compared by intra-class correlation coefficients (t), calculated from a one-way ANOVA with individual as factor (Lessels & Boag1987; Boake 1989; Sokal & Rohlf 1995). An intraclasscorrelation coefficient close to 1 indicates a high repeat-ability, one close to zero none. We examined the repeat-ability at longer intervals, that is, with variable periods offood stress, in 20 spiders kept at 20�C during a 3-weekexperiment, in which they were fed only on the firstday of each week. Consequently, they experiencedincreasing food deprivation effects for the rest of theweek. Ballooning latency was tested each day. Intraclasscorrelation coefficients for ballooning latency were calcu-lated between days when spiders were assumed to be inthe same nutritional state (day after spiders were fed,1-week intervals). All analyses were performed with SAS8.1 and statistica (Statsoft 2000); variation caused by therandom individual effect (intraclass correlation) wastested against a model including all variance components(chi-square test of difference in log likelihood). Statisticaltests were two tailed.

Developmental Conditions and Family

We randomly assigned spiderlings from 13 females tothe four rearing conditions described above, and tested

171BONTE ET AL.: SPIDER BALLOONING BEHAVIOUR

them 1 day after final moult under well-fed conditions.Preliminary experiments indicated that gender did notinfluence ballooning latency (unpublished data). Weapplied a General Linear Model with type III sum ofsquares, family (genotype) as a random factor and tem-perature and feeding conditions as fixed factors. In thisway, our data were analysed in a quantitative geneticframework with genotype�environment interaction,whereby genotype included nonadditive components,prenatal maternal and paternal effects. We used 284individual spiders from 13 families (mean 5.04 siblings/treatment). Variance estimates, contributed by specialenvironmental effects (as a consequence of randomdevelopmental noise and microenvironmental variation),genotype and genotype�environment interactions werecalculated from the expected mean squares under amixed model (Lynch & Walsh 1998). Because of thespecific mean squares configuration, fixed (macro-environmental) factors did not explain any of thevariation (Lynch & Walsh 1998).

Relations with Life History Traits

We measured development time to maturity (to thenearest day) and size of the spider (cephalothorax size inmm2 digitized under a 50� binocular microscope) forthe offspring of 24 females reared with prey ad libitum at20�C. These data were used for the analysis of possiblecorrelations with the ballooning latency. In addition, wemeasured female size (measured in the same way as forthe offspring) and the number of offspring in thefirst cocoon (fecundity as maternal fitness characteristic).The sex ratio in the first cocoon was also determined.Because fecundity was correlated with the size of thefemale (Pearson correlation: r148=0.268, P<0.05), we usedresiduals of the regression of number of eggs in the firstcocoon on female size as a fitness measure.

RESULTS

Repeatability and Acute Stress

Ballooning latency did not differ significantly betweenthe two tests conducted within 1 h and was highlyrepeatable (X�SD; test 1: 110.66�66.90 s; test 2:87.94�50.91 s; intraclass correlation t=0.781; randomeffect: �2

1=43.5, P<0.001). At 1-week intervals it decreasedand was less repeatable (range 0.306–0.708). Ballooninglatency decreased significantly over the first week ofthe 3-week experiment (Pearson correlation: r18=0.270,P<0.05) but not during the second (r18=0.014, NS) andthe third weeks (r18=0.0008, NS; Fig. 1).

Developmental Conditions and Family

Phenotypic variation in ballooning latency could beattributed to feeding and temperature conditions duringjuvenile development, but not to genotypic (family)effects (Table 1). Levels of ballooning latency were signifi-cantly higher at 15�C than at 20�C and higher with the

poor diet (Fig. 2). Latency differed significantly betweentemperatures (X�SE; 15�C: 4.979�0.783 s; 20�C:5.493�0.541 s; post hoc Scheffe test: P<0.001) but notbetween diet treatments (poorly fed: 5.251�0.625 s; wellfed: 5.320�0.674 s; post hoc Scheffe test: NS). The inter-action between temperature and diet was, however, sig-nificant: when poorly fed, spiderlings reared at 20�Cincreased ballooning latencies, and those reared at 15�Cdecreased it.

Genetic variation explained 29.2% of the total pheno-typic variation (partitioned as 16.0% for genotype and13.2% for all genotype�environment interactions);the remaining variation could be attributed to micro-environmental variation (70.8%; Table 1).

Relations with Life History Traits

For males and females separately, no relation betweenballooning latency and size at maturity was found (maleswere significantly larger than females: t test: t236=10.45,P<0.001; Table 2). Development time to maturity wasindependent of sex (t278=0.116, NS) and did not show asignificant correlation with ballooning latency for thewell-fed individuals reared at 20�C (Table 2). If we includeall observations from the four-environments experiment,a significant positive correlation was found with develop-ment time increasing in the order high temperature/wellfed, high temperature/poorly fed, low temperature/wellfed and low temperature/poorly fed. Both temperatureand diet contributed to this difference (ANOVA: tempera-ture effect: F1,280=354.921, P<0.001; diet effect: F1,280=33.927, P<0.001). No significant relations with maternallife history traits were found (Table 2).

020

600

Day

Bal

loon

ing

late

ncy

(s)

0

500

400

300

200

100

5 10 15

Week 1 Week 2 Week 3

Figure 1. Ballooning latency (X±95% confidence intervals, s) over a3-week experiment in which 20 spiders were fed only on the first dayof the week (arrow).

DISCUSSION

Our laboratory experiments showed high repeatability oflatency to express preballooning behaviour under stress(when the spider had to escape from the test chamber tosurvive at a relatively high temperature) at short intervals

172 ANIMAL BEHAVIOUR, 66, 1

Table 1. Results of the GLM for ballooning latency (ln transformed) on the factors temperature, diet (both fixed)and family (random)

Factor Sum of squares dfMeansquare F P Variance estimates

Temperature 8.456 1 8.456 8.222 0.012 NEDiet 1.927 1 1.927 5.159 0.036 NEFamily 30.948 12 2.579 3.720 0.159 0.105Temperature×Diet 3.201 1 3.201 4.327 0.032 NETemperature×Family 13.413 12 1.117 1.426 0.273 0.065Diet×Family 4.308 12 0.359 0.458 0.904 −0.010Temperature×Diet×Family 9.402 12 0.784 1.648 0.070 0.032Error 121.424 261 0.465 0.465

NE: No estimates available.

4.6Poorly fed

5.6

Ln b

allo

onin

g la

ten

cy (

s) 5.4

5.2

5

4.8

Well fed

15°C20°C

Figure 2. Ballooning latency (X±SE, s) of spiders that, as juveniles,were reared at 15 or 20°C and were either well fed or poorly fed(four prey per 3 or 7 days).

of 1 h, but low repeatability at longer intervals ofa week. Phenotypic variation was determined mainly byenvironmental conditions and less so by genotypiceffects. Relations with individual and maternal fitnesscharacteristics were not found.

Our results confirm that tiptoe behaviour in erigonidspiders is strongly influenced by food deprivation (VanWingerden & Vughts 1974; Legel & Van Wingerden1980; Weyman & Jepson 1994; Weyman et al. 1994). Thispattern was found only during the first week of fooddeprivation, however, and not during food deprivation inthe next 2 weeks. This result indicates that ballooninglatency depends not only on acute feeding stress but alsoon feeding history and age. Because of this low repeat-ability, it is difficult to make representative multiplemeasurements and to estimate the trait heritability underdifferent environmental and internal conditions. Thelack of multiple measurements has led to underestima-tion of the levels of heritability of courtship and matingspeed in Drosophila (Hoffmann 1999). The male-biasedsex ratio during ballooning in the field (Blandenier &Furst 1998; Bonte et al. 1998) cannot be attributed to adifferent tiptoe-initiating behaviour, because the latterwas independent of sex.

Different environmental conditions during juveniledevelopment markedly affected the latency to initiate

preballooning behaviour. Both the feeding regime andthe temperature had a significant effect. The effect oftemperature was clear: spiders reared at 20�C took offfaster than those reared at 15�C. The finding that thejuvenile diet influenced the adult’s ballooning latencywas as predicted; however, our results showed an effect oftemperature as well. Ballooning latency was shorter forspiders that were well fed during development when thetemperature was high (20�C), but at the lower breedingtemperature, there was only a marginally significant dif-ference between diets. An interaction between diet andtemperature was thus present, indicating a complexreaction of spiders to different postnatal environmentalfactors.

The strong effect of temperature during rearing onballooning latency may be an inherent physiologicalresponse. Flight activity in Oncopeltus also changes signifi-cantly with temperature during development (Dingle1968), suggesting a physiological depression near tem-peratures at the threshold where reproduction is possible.In the carabid beetle Pogonus chalceus, where dispersalpotential is related to wing size, temperature duringdevelopment did not alter the relative wing length(Desender 1989). High temperatures over the summercould thus act as an extra trigger for the stimulation ofthe mass late-summer ballooning commonly observed inE. atra (Bonte et al. 1998; Duffey 1998), although aninteraction with temperature during ballooning is to beexpected. Nevertheless, our data indicate that ballooninglatency is influenced by physiological responses toenvironmental circumstances during juvenile develop-ment. Mothers are thus able to influence the ballooninginitiation of their offspring by the choice of habitat forreproduction.

Different nutritional conditions during juvenile devel-opment induced different behavioural ballooning pat-terns, in contrast to acute food stress, which is the finaltrigger for ballooning in many spiders (Van Wingerden &Vughts 1974; Legel & Van Wingerden 1980; Weyman &Jepson 1994; Weyman et al. 1994; Bonte & Maelfait2001). Our findings contradict those of Legel & VanWingerden (1980), who found a lower percentage ofballooners in tested spiders when individuals were rearedunder well-fed and medium-starved regimes (40 and37.5%) than with greater levels of starvation. Their model

173BONTE ET AL.: SPIDER BALLOONING BEHAVIOUR

organism was Erigone arctica (White, 1852), a sister speciestypical of salt marshes. Although the power of theirexperiment was too low for appropriate statistics, theirresults indicate again that feeding conditions duringdevelopment alter ballooning capacity. Different speciescan apparently react in different ways to changingenvironmental conditions. Since the temperature duringrearing in their experiment was rather high (room tem-perature between 18 and 23�C), similar results to ourscould be expected.

Our results suggest that ballooning latencies in E. atraare short only in highly productive habitats, where preyare abundant and temperatures high. As we observedcovariance between development time in the four treat-ments and ballooning latency, the variation we observedin ballooning latency can be explained by variation indevelopment time. In our experiment, however, balloon-ing latency was not correlated with development timewithin the separate rearing conditions and so develop-ment time is probably not the main trigger responsiblefor the observed variation in ballooning latency underlaboratory conditions.

Family effects explained about 16% of the observedphenotypic variation. Although not statistically signifi-cant (only 13 families were included in the quantitativestudy), this genotypic source of variation was notnegligible. It stresses the importance and interaction ofgenetic components in the expression of behaviouraltraits, narrowly linked to dispersal. This biologicallyrelevant, low genotypic source of variation is concordantwith Roff & Mousseau’s (1987) and Hoffmann’s (1999)findings of overall low heritabilities for behavioural traits.We did not find any relation between offspring balloon-ing latency and maternal fitness characteristics. The prob-ability of ballooning by young crab spiders, Misumenavatia, also did not differ with clutch size or individual size(Morse 1993). Morse (1993) also found an effect of sub-strate, which was related to immediate prey availability.Consequently, the choice of nest sites (substrate) by thespiderlings’ mothers indirectly affected the spiderlings’

dispersal behaviour. Our results are thus in agreementwith Morse’s (1993) findings that intraspecific variationin spider ballooning is only marginally determined byevolutionary components. More empirical research isneeded to understand the possible sources of variation inballooning behaviour.

In insects with distinct dispersal characteristics such aswing polymorphisms and differentiated muscle develop-ment, maternal and additive genetic effects influence thetotal phenotypic variation (e.g. Janssen et al. 1988; Leamy1988; Desender 1989). In full and half sibling, direct andindirect selection experiments, Li & Margolies (1993,1994) found a relatively high heritability of ballooninglatency in the arachnid (phytophagic) mite Tetranychusurticae. Li & Margolies minimized the residual varianceduring the experiments to 43–60%; and they estimatedthe additive genetic component to be 22.39%. Because ofthe strong influence of environmental factors and theimpossibility of conducting multiple measurements,analogous experiments on ballooning behaviour inspiders are probably difficult to conduct in a standardizedway.

In summary, we conclude that, in addition to (exten-sively documented) acute starvation, environmental vari-ation during juvenile development influences ballooninglatency in E. atra under our laboratory conditions. Thehabitat choice of the mother is therefore likely to influ-ence offspring ballooning behaviour. A low, but biologi-cally relevant, genotypic source of variation was,however, detected indicating that environment andgenotype sum and interact in the expression of thisdispersal-linked behavioural trait.

Table 2. Life history traits of offspring and their mother, and correlations between traits and ballooning latency

Life history trait Mean value±SD N

Correlation

r* P

OffspringDevelopment time (days)

All conditions 39.23±5.18 280 0.220 <0.001At 20°C, spiders well fed 22.05±3.34 248 0.067 NS

Cephalothorax size at maturity (mm2)Males 3.34±0.21 117 0.142 NSFemales 3.09±0.16 125 0.042 NS

MotherOffspring sex ratio (males/females) 0.87±0.38 24 −0.129 NSAdult cephalothorax size (mm2) 3.20±0.03 24 −0.259 NSNumber of offspring in first cocoon 14.00±3.95 24 0.146 NSResiduals of offspring on mother’s size — 24 −0.306 NS

*Pearson correlations for offspring traits and Spearman rank correlations for mother’s traits.

Acknowledgments

We thank L. Maes for measuring the cephalothoraxwidths of the E. atra offspring. Luc Lens, MarkAlderweireldt and three anonymous referees gave veryvaluable comments on the manuscript.

174 ANIMAL BEHAVIOUR, 66, 1

References

Blandenier, G. & Furst, P. A. 1998. Ballooning spiders caughtby a suction trap in an agricultural landscape in Switzerland.Proceedings of the 17th European Colloquium of Arachnology,Edinburgh, 177–185.

Boake, C. R. B. 1989. Repeatability: its role in evolutionary studies ofmating behavior. Evolutionary Ecology, 3, 173–182.

Bonte, D. & Maelfait, J.-P. 2001. Life history, habitat use anddispersal of a dune wolf spider (Pardosa monticola (Araneae:Lycosidae)) in the Flemish coastal dunes (Belgium). Belgian Journalof Zoology, 131, 141–153.

Bonte, D., Hoffmann, M. & Maelfait, J.-P. 1998. Resultatenvan een onderzoek naar aeronautisch actieve spinnen in hetduingebied van Oostduinkerke (West-Vlaanderen). Nieuwsbriefvan de Belgische Arachnologische Vereniging, 13, 77–85.

De Keer, R. & Maelfait, J.-P. 1988a. Laboratory observations on thedevelopment and reproduction of Erigone atra Blackwall, 1833(Araneae, Linyphiidae). Bulletin of the British Arachnological Society,7, 237–242.

De Keer, R. & Maelfait, J.-P. 1988b. Observations on the life cycleof Erigone atra (Araneae, Erigoninae) in a heavily grazed pasture.Pedobiologia, 32, 201–212.

Dean, D. A. & Sterling, W. L. 1985. Size and phenology ofballooning spiders at two locations in eastern Texas. Journal ofArachnology, 13, 111–120.

Desender, K. 1989. Heritability of wing development and bodysize in a carabid beetle, Pogonus chalceus Marsham, and itsevolutionary significance. Oecologia, 78, 513–520.

Dingle, H. 1968. The influence of environment and heredityon flight activity in the milkweed bug Oncopeltus. Journal ofExperimental Biology, 48, 175–184.

Duffey, E. 1998. Aerial dispersal in spiders. Proceedings of the 17thEuropean Colloquium of Arachnology, Edinburgh, 187–191.

Greenstone, M. H. 1982. Ballooning frequency and habitatpredictability in two wolf spiders (Lycosidae: Pardosa). CanadianEntomologist, 65, 84–89.

Greenstone, M. H. 1990. Meteorological determinants of spiderballooning: the roles of thermals vs. the vertical wind speedgradient in becoming airborne. Oecologia, 84, 164–168.

Greenstone, M. H., Morgan, C. E. & Hultsch, A.-L. 1985.Ballooning methodology: equations for estimating masses ofsticky-trapped spiders. Journal of Arachnology, 15, 225–230.

Greenstone, M. H., Morgan, C. E. & Hultsch, A.-L. 1987.Ballooning spiders in Missouri, USA, and New South Wales,Australia: family and mass distributions. Journal of Arachnology, 15,163–170.

Hoffmann, A. A. 1999. Is the heritability for courtship and matingspeed in Drosophila (fruit fly) low? Heredity, 82, 158–162.

Janssen, G. M., De Long, G., Joose, E. N. G. & Scharloo, W. 1988.A negative maternal effect in springtails. Evolution, 42, 828–834.

Jarvinen, O. & Vepsalainen, K. 1976. Wing dimorphism as anadaptive strategy in water-striders (Gerris). Hereditas, 84, 61–68.

Jones, D. 1996a. How ballooners become airborne. Newsletter of theBritish Arachnological Society, 69, 5–6.

Jones, D. 1996b. How ballooners become airborne: a postscript.Newsletter of the British Arachnological Society, 70, 4.

Kaitala, A. 1988. Wing muscle dimorphism: two reproductivepathways of the waterstrider Gerris thoracicus in relation to habitatinstability. Oikos, 53, 222–228.

Kecic, V. & Marinkov, D. 1974. Multiple choice selection for lightpreference in Drosophila subobscura. Behavioral Genetics, 4, 285–300.

Kirkpatrick, M. & Lande, R. 1989. The evolution of maternalcharacters. Evolution, 43, 485–503.

Leamy, L. 1988. Genetic and maternal influences on brain and bodysize in random bred house mice. Evolution, 42, 42–53.

Legel, G. J. & Van Wingerden, W. K. R. E. 1980. Experiments onthe influence of food and crowding on the aeronautic dispersal ofErigone arctica (White, 1852) (Araneae, Linyphiidae). Proceedingsof the 8th International Congress of Arachnology, Vienna, 97–103.

Lessels, C. M. & Boag, P. T. 1987. Unrepeatable repeatabilities: acommon mistake. Auk, 104, 116–121.

Li, J. & Margolies, D. C. 1993. Quantitative genetics of aerialdispersal behaviour and life-history traits in Tetranychus urticae.Heredity, 70, 544–552.

Li, J. & Margolies, D. C. 1994. Responses to direct and indirectselection on aerial dispersal behaviour in Tetranychus urticae.Heredity, 72, 10–22.

Lynch, M. & Walsh, B. 1998. Genetics and Analysis of QuantitativeTraits. Sunderland, Massachusetts: Sinauer.

Massot, M. & Clobert, J. 2000. Processes at the origin of similaritiesin dispersal behaviour among siblings. Journal of EvolutionaryBiology, 13, 707–719.

Miller, G. L. 1984. Ballooning in Geolycosa turricola (Treat) andGeolycosa patellonigra Wallace: high dispersal frequencies in stablehabitats. Canadian Journal of Zoology, 62, 2110–2111.

Morse, D. H. 1993. Some determinants of dispersal by crabspiderlings. Ecology, 74, 427–432.

Richter, C. J. 1970. Aerial dispersal in relation to habitat in eight wolfspider species (Pardosa, Araneae, Lycosidae). Oecologia, 5, 200–214.

Richter, C. J. 1971. Some aspects of aerial dispersal in differentpopulations of wolf spiders, with particular reference to Pardosaamentata (Araneae, Lycosidae). Miscellaneous Papers LandbouwHogeschcool, Wageningen, 8, 77–88.

Roff, D. A. 1986. The evolution of wing dimorphism in insects.Evolution, 40, 1009–1020.

Roff, D. A. & Mousseau, T. A. 1987. Quantitative genetics andfitness: lessons from Drosophila. Heredity, 58, 103–118.

Sokal, R. R. & Rohlf, F. J. 1995. Biometry. 3rd edn. San Francisco:W. H. Freeman.

Statsoft 2000. Statistica for Windows. Tulsa: Statsoft.Thomas, C. F. G. 1996. Modeling aerial dispersal of Linyphiid

spiders. Aspects of Applied Biology, 46, 217–222.Van Wingerden, W. K. R. E. 1980. Aeronautic dispersal of

immatures of two Linyphiid spider species (Araneae, Linyphiidae).Proceedings of the 8th International Congress of Arachnology.Vienna, 91–96.

Van Wingerden, W. K. R. E. & Vugts, H. F. 1974. Factorsinfluencing aeronautic behaviour of spiders. Bulletin of the BritishArachnological Society, 3, 6–10.

Vugts, H. F. & Van Wingerden, W. K. R. E. 1976. Meteorologicalaspects of aeronautic behaviour of spiders. Oikos, 27, 433–444.

Weyman, G. S. 1993. A review of the possible causative factorsand significance of ballooning in spiders. Ethology, Ecology andEvolution, 5, 279–291.

Weyman, G. S. & Jepson, P. C. 1994. The effect of food supply onthe colonisation of barley by aerially dispersing spiders (Araneae).Oecologia, 100, 386–390.

Weyman, G. S., Sunderland, K. D. & Fenlon, J. S. 1994. The effectof food deprivation on aeronautic dispersal behaviour (ballooning)in Erigone spp. spiders. Entomologia Experimentalis et Applicata, 73,121–126.

Weyman, G. S., Jepson, P. C. & Sunderland, K. D. 1995. Doseasonal changes in numbers of aerially dispersing spiders reflectpopulation density on the ground or variation in ballooningmotivation? Oecologia, 101, 487–493.